In a network operation management system, a communication monitor that predicts an alarm occurring in a network has been known. The communication monitor records an occurrence time of an alarm, a type of an alarm, and the like which are detected by a network device in a network as a change history, and predicts an alarm that will occur next using the change history.
For example, it is assumed that the communication monitor has a record representing that the frequency of occurrence of a pattern A in which an alarm 2 occurs after an alarm 1 occurs is 5 times and that the frequency of occurrence of a pattern B in which an alarm 3 occurs after the alarm 1 occurs is 4 times as the change history. When the occurrence of the alarm 1 is detected, the communication monitor searches the change history for a pattern using the alarm 1, and then determines the pattern A which is high in the frequency of occurrence as a search result. As a result, when the pattern A is determined, the communication monitor predicts the alarm 2 as an alarm that will occur after the alarm 1 occurs. Then, the communication monitor notifies the user of prediction of the occurrence of the alarm 2.
The communication monitor predicts an alarm of a “congestion failure” occurring as a traffic volume dynamically varies as well. However, in the congestion failure, for example, there are cases in which as a path is switched immediately after a congestion failure 1 occurs, a traffic volume of a path used after switching varies, and so another congestion failure 2 occurs. In this case, switching to another path is performed after the congestion failure 2 occurs. Then, for example, the communication monitor keeps a record representing that the frequency of occurrence of the pattern A in which the congestion failure 2 occurs after the congestion failure 1 occurs is 5 times and the frequency of occurrence of the pattern B in which the congestion failure 3 occurs after the congestion failure 1 occurs is 4 times as the change history.
Then, when the congestion failure 1 occurs, the communication monitor predicts the occurrence of the congestion failure 2 according to the pattern A which is high in the frequency of occurrence based on the change history such as the frequency of occurrence.    Patent Document 1: Japanese Laid-open Patent Publication No. 2009-10438    Patent Document 2: Japanese Laid-open Patent Publication No. 2005-285040
Since a congestion failure is a failure that occurs according to a traffic volume of data flowing through a path in a network, for example, there are cases in which the congestion failure 2 does not occur in practical even when the congestion failure 1 occurs. However, for example, when the congestion failure 1 occurs, the communication monitor simply predicts an alarm of the occurrence of the congestion failure 2 based on the change history such as the frequency of occurrence in the same way.